Content uploaded by Geoffrey E. Hill
Author content
All content in this area was uploaded by Geoffrey E. Hill on Nov 23, 2015
Content may be subject to copyright.
The Importance of Carotenoid Dose in Supplementation
Studies with Songbirds
Rebecca E. Koch
1,
*
Alan E. Wilson
1,2
Geoffrey E. Hill
1
1
Department of Biological Sciences, 331 Funchess Hall,
Auburn University, Auburn, Alabama 36849;
2
School of
Fisheries, Aquaculture, and Aquatic Sciences, Auburn
University, Auburn, Alabama 36849
Accepted 10/9/2015; Electronically Published 11/16/2015
ABSTRACT
Carotenoid coloration is the one of the most frequently studied
ornamental traits in animals. Many studies of carotenoid col-
oration test the associations between carotenoid coloration and
measures of performance, such as immunocompetence and
oxidative state, proceeding from the premise that carotenoids
are limited resources. Such studies commonly involve supple-
menting the diets of captive birds with carotenoids. In many
cases, however, the amount of carotenoid administered is poorly
justified, and even supposedly carotenoid-limited diets may sat-
urate birds’systems. To quantify the relationships among the
amount of carotenoids administered in experiments, levels of
circulating carotenoids, and quantities of carotenoids deposited
into colored ornaments, we performed a meta-analysis of 15
published studies that supplemented carotenoids to one of seven
songbird species. We used allometric scaling equations to esti-
mate the per-gram carotenoid consumption of each study’s sub-
jects, and we used meta-regression to evaluate the effects of this
carotenoid dose on differences in coloration and plasma carot-
enoid levels between supplemented and control groups of birds.
After accounting for supplementation duration and species, we
observed a significant positive correlation between carotenoid
intake and response of plasma carotenoid level to supplementa-
tion. The presence of supplemental carotenoids also tended to
increase the expression of ornamental coloration, but the mag-
nitude of the carotenoid dose did not significantly affect how
strongly coloration changed with supplementation. Further, col-
oration effect sizes had no significant relationship with plasma
carotenoid effect sizes. We also found significant heterogeneity
in responses among studies and species, and the parameters used
to measure color significantly affected response to supplementa-
tion. Our results emphasize the importance of performing dos-
age trials to determine what supplementation levels provide
limited versus surplus carotenoids and of measuring the natural
level of carotenoid intake by the study species to validate the
appropriateness of supplementation levels for a particular study
species and experimental design.
Keywords: carotenoid supplement, ornamental coloration,
plasma carotenoids, plumage coloration, bill coloration.
Introduction
Carotenoid-based ornaments in birds have drawn substantial
attention as indicator traits because numerous studies have
reported correlations between the expression of carotenoid
coloration and aspects of male quality, including fat reserves,
basal metabolic rate, effectiveness of immune response (im-
munocompetence), and oxidative state (reviewed in Hill 2002,
2006; Svensson and Wong 2011). Carotenoid pigments are re-
sponsible for most of the vibrant red, orange, and yellow color-
ation of the feathers and soft parts of birds (McGraw 2006),
and they may also play important physiological roles as vita-
min A precursors, boosters of the immune system, and antioxi-
dants (Mougeot et al. 2010; Pérez-Rodríguez et al. 2010; Hill and
Johnson 2012). Because these pigments cannot be synthesized
in the bodies of animals and must be acquired from the diet
(Goodwin 1984), carotenoids are often considered limited re-
sources such that only birds in the best condition can afford to
allocate carotenoid pigments toward colored ornaments rather
than retain them for potential internal benefit; thus, carotenoid
resource trade-offs have been hypothesized to maintain signal
honesty in these traits (Møller et al. 2000; Alonso-Alvarez et al.
2004).
Numerous studies of carotenoid ornamentation aim to es-
tablish and clarify whether this hypothesized carotenoid re-
source trade-off may explain the condition dependence of ca-
rotenoid coloration in birds by validating that (1) higher levels of
circulating carotenoids improve immune function and/or oxi-
dative stress maintenance, (2) restricted dietary intake limits the
quantity of circulating carotenoids, and (3) generation of a high-
quality ornament sequesters circulating carotenoids such that
colored traits impose a cost on other processes that utilize ca-
rotenoids (von Schantz et al. 1999; Møller et al. 2000; Alonso-
Alvarez et al. 2008). Fundamental to testing these predictions of
the trade-off hypothesis are experiments that manipulate carot-
enoid availability and measure the effect of carotenoid dose on
*Corresponding author; e-mail: rek0005@auburn.edu.
Physiological and Biochemical Zoology 89(1):000–000. 2016. q2015 by The
University of Chicago. All rights reserved. 1522-2152/2016/8901-5081$15.00.
DOI: 10.1086/684485
000
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
both ornamentation and physiology. In laboratory settings, re-
searchers commonly supplement or restrict dietary carotenoid
levels and evaluate the resulting effects on various measures of
ornamentation and internal condition (Hill 2006). However,
the results of these studies are often inconclusive, and the im-
portance of allocation trade-offs to carotenoid-based signal hon-
esty as well as the physiological functions of carotenoid them-
selves remain debated (Hill 1994, 1999, 2011, 2014; Hudon 1994;
Hartley and Kennedy 2004; Hadfield and Owens 2006; Costan-
tini and Møller 2008).
One critical but often overlooked complication of carotenoid
manipulation studies is the biological relevance of the quan-
tities of carotenoids that are administered to test animals. Com-
monly, supplemental carotenoids are provided ad lib. in food
or water without an assessment of the amounts of pigments that
are actually ingested and without proper consideration for how
levels of supplemental carotenoids compare with quantities in-
gested by birds under natural conditions. Moreover, the quan-
titative relationship between the amount of carotenoids ingested
and quantities of circulating carotenoids is usually not measured
in either lab or field systems, so it is difficult to judge the results
of carotenoid supplementation. For example, the quantity of in-
gested carotenoids may greatly exceed that which is present in
the plasma if birds rapidly transport consumed carotenoids to
storage in fat, ornamentation, or other tissues; therefore, birds
with vastly different carotenoid access may have the same levels
of plasma carotenoids if the bird with greater consumption al-
locates his excess carotenoids outside of circulation. For this rea-
son, comparing plasma carotenoid levels of captive birds to wild
conspecifics is insufficient to justify that the captive supple-
mentation dose mimics the levels of carotenoids available to
wild birds. Because the differential allocation of limited carot-
enoids is key to the resource trade-off hypothesis, it is essential
to better track carotenoid usage through quantifying the re-
lationships among amounts ingested, circulated, and deposited
in ornaments.
Several studies have addressed this issue by using dosage
trials to compare supplementation levels to levels of circulating
carotenoids in order to identify doses that do not saturate their
subjects’systems (e.g., Alonso-Alvarez et al. 2004; Aguilera and
Amat 2007). Too often, however, the carotenoid supplemen-
tation regimens used in avian studies are based on methods
developed for other species or from studies of different dietary
carotenoids (e.g., Navara and Hill 2003; Baeta et al. 2008);
carotenoid consumption and absorption varies markedly across
species with different masses and life histories (Tella et al. 2004;
McGraw 2005), so extrapolating carotenoid doses among spe-
cies with no validation could lead to experiments that provide
carotenoid doses that are too high or too low to yield meaningful
results.
Because the focus of most studies utilizing carotenoid sup-
plementation is testing for trade-offs in the use of limited ca-
rotenoid resources for ornamentation versus body maintenance,
poorly controlled dosing undermines the goals of the research.
For a study of resource limitation or trade-off to be meaningful,
then the resource must be provided at a level below saturation.
If the lowest supplementation level provides sufficient carot-
enoids for both body maintenance and ornament production,
then studying the effects of dose becomes meaningless. As fun-
damental as these ideas appear to be, many studies proceed on
the unstated assumption that supplementation levels are below
saturation.
To better quantify the effects of supplementation on circu-
lating carotenoid availability and carotenoid-based ornamen-
tation, we performed a meta-analysis of 15 published studies
that include groups of both carotenoid supplemented and
unsupplemented birds and that report the resulting plasma
carotenoid levels and/or ornamental color of each group. A
previous meta-analysis investigated correlations among these
variables, but it grouped studies as either supplemented or
unsupplemented without including supplementation dose as
a cofactor (Simons et al. 2012), missing a critical source of var-
iation among studies. In our analysis, we built on the existing
literature by first using published levels of carotenoid supple-
mentation and allometric scaling equations to estimate indi-
vidual consumption of carotenoids. We then modeled how
variation in intake between supplemented and control groups
affected the relationships between circulating carotenoids and
allocation to ornamentation in songbirds. By quantifying the
physiological responses to varying levels of carotenoid inges-
tion in different studies and seven different songbird species,
we provide a foundational model for predicting the biological
relevance of particular carotenoid supplementation regimens
and can assess the variables that modulate response to carot-
enoid intake.
Methods
Literature Search
We surveyed the existing carotenoid literature using the Web of
Science database on March 23, 2014, using the keywords “ca-
rotenoid*”AND “supp*”AND “bird”OR “avian.”We included
only studies (1) reporting the level of carotenoid supplemen-
tation as well as the food source provided; (2) including data on
both carotenoid-supplemented and control groups of indi-
viduals; (3) reporting the values of plasma carotenoid levels
and/or coloration; (4) not repeating measures on the same
group of birds that were reported in a study already incorpo-
rated into the meta-analysis (a potential source of pseudorep-
lication); (5) testing adult male birds rather than nestlings (in
which both carotenoid physiology and ornamental function
differ greatly from sexually reproducing adult birds, and the
quantity of carotenoids acquired from egg yolk or parental
provisioning is often unknown; Hill and McGraw 2006); and
(6) supplementing with only the carotenoids lutein and/or zea-
xanthin, the most prevalent carotenoid pigments in the avian
diet (McGraw 2006). With the exception of one study sup-
plementing with only lutein (Stirnemann et al. 2009), all stud-
ies included in our meta-analysis supplemented primarily with
lutein and trace amounts of zeaxanthin (e.g., 20∶1 ratio of lu-
tein∶zeaxanthin; Blount et al. 2003; Hõrak et al. 2007; Karu et al.
2007; Baeta et al. 2008; Sild et al. 2011; Sepp et al. 2011).
000 R.E.Koch,A.E.Wilson,andG.E.Hill
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
This latter point is important because most terrestrial birds
consume diets containing primarily these two yellow and
structurally similar carotenoid pigments, which many species
must then metabolize into red pigments (in species with red
coloration) or ornamental yellow pigments (e.g., canary xan-
thophylls). Critically, chemical properties and therefore po-
tential physiological functions vary across these dietary and
ornamental pigments, and the costs of converting dietary to
ornamental pigments may play a key role in the honesty of
carotenoid-based coloration (Hill 1996; Hill and Johnson 2012;
Johnson and Hill 2013). Studies supplementing with other
pigments, particularly the red carotenoids at the end points of
these carotenoid conversion pathways (such as canthaxanthin;
e.g.,McGrawetal.2002;Smithetal.2007),bypasssomeofthe
mechanisms relating coloration to physiology that may be
important to carotenoid signal honesty, so such studies are not
appropriate for this analysis.
Despite the extensive literature on carotenoid ornamenta-
tion (more than 300 results to our initial keyword search), only
19 studies met our criteria of providing measurable caroten-
oid supplementation quantities to adult birds. Because 16 of
19 studies investigated songbird species (order Passeriformes),
we excluded one study of red junglefowl (Gallus gallus;McGraw
and Klasing 2006), one study of mallards (Anas platyrhynchos;
Butler and McGraw 2013), and one study of kestrels (Falco
tinnunculus; Costantini et al. 2007) to capture the majority
of available data while avoiding comparing data from phy-
logenetically distant taxa with different physiologies. We
also excluded one study on society finches (Lonchura striata
domestica; McGraw et al. 2006) because this species lacks
carotenoid-based ornamentation and so is not subject to the
potential costs of allocating carotenoids as colorants. We per-
formed our analysis on the remaining 15 studies of seven
songbird species with carotenoid-based ornaments: the Ameri-
can goldfinch (Carduelis tristis) with yellow plumage and
pink-red bill ornamentation, the house finch (Haemorrhous
mexicanus) with red plumage ornamentation, the zebra finch
(Taeniopygia guttata) with red bill ornamentation, the diamond
firetail (Stagonopleura guttata) with red bill ornamentation, the
great tit (Parus major) with yellow plumage ornamentation, the
Eurasian blackbird (Turdus merula) with red-orange bill orna-
mentation, and the European greenfinch (Carduelis chloris) with
yellow plumage ornamentation.
Carotenoid Supplementation Calculations
Most experiments supplemented carotenoids to the main food
orwatersupplyandreporteddosesastheconcentrationof
carotenoids added per unit food or water. One study by Peters
et al. (2011) quantified daily carotenoid intake of individuals
during the experiment, so these values were used in our anal-
ysis. For all other studies, we estimated the quantity of ca-
rotenoids consumed by each bird by first calculating the av-
erage daily food or water intake of an individual of the focal
species, using allometric scaling equations to account for the
nonlinear relationship between species size and consumption.
When carotenoids were supplemented in the water supply,
we estimated daily water intake using the mass of the study
species and the scaling equation for passerines reported by
Calder and Braun (1983). When a study supplemented ca-
rotenoids in the food supply, we estimated the daily food in-
take of the study’s focal species by using the energy content of
the food provided (often, millet or sunflower seeds; Caraco et al.
1980; Hõrak et al. 2003) and a scaling equation for passerines
that predicts the consumption needed to meet daily energetic
requirements (Nagy et al. 1999). When the exact mass of in-
dividuals included in the study was not reported, we estimated
the average mass of the species from the Handbook of the Birds
of the World (del Hoyo 2010). From our estimates of daily food
or water intake, we then used each study’spublisheddetailson
the concentration of carotenoids supplemented to calculate
the quantity of carotenoids ingested along with food or water.
We also calculated the carotenoid content of the basic diet
provided to both control and supplemented birds, using re-
ported carotenoid content values or published measurements
of the content of the seeds supplied (McGraw et al. 2001; Peters
et al. 2008) to account for dietary carotenoids acquired inde-
pendently of supplementation (app. A, available online).
To standardize levels of supplemental carotenoids ingested
in species of varying body sizes, we divided daily carotenoid
consumption amount by species mass in grams. We then
calculated the difference in carotenoid intake between sup-
plemented and control groups for each study (carotenoid intake
difference). Most often, this measure of intake difference was
nearly identical to the actual intake of the supplemented group,
since most control groups acquired negligible levels of carot-
enoids.
Effect Size Calculations
We calculated the natural log response ratio and its variance
from reported means and standard deviations of control and
supplemented groups according to the formulas outlined by
Koricheva et al. (2013); the response ratio allows for the stan-
dardization of measurements across studies by converting each
measured effect into a unitless ratio of the mean response of
the supplemented group to the mean response of the control
group. When other experimental manipulations were present
in a study, we used data from the otherwise unaltered control
groups that varied only in carotenoid supplementation. We cal-
culated two types of effect sizes per study, when possible, to
measure the effects of supplementation on plasma carotenoid
levels and ornamental coloration. When necessary, we ex-
tracted means and errors from figures using either ImageJ
(Rasband 1997–2014) or WebPlotDigitizer v. 2.6 (Rohatgi
2013). When mean values were not published in text or fig-
ures, we contacted authors to retrieve the raw data and calculate
mean values. Along with effect size, we recorded each study’s
focal species and the number of days that supplementation was
provided. If a study reported multiple response values over time,
we recorded only values from before supplementation and at
the end of supplementation for consistency among studies.
Importance of Carotenoid Dose in Songbird Supplementation 000
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
We included multiple effect sizes for one study only if each
differed in a key variable, such as a different carotenoid sup-
plementation dose or ornament measured. In addition, because
the color of feathers is determined only during molt when
carotenoids are actively deposited in growing feathers (Hill 2002),
we extracted plumage color effect sizes only from studies of
molting individuals; we calculated effect sizes from nonmolting
birds with plumage ornaments only for the relationship between
carotenoid intake and plasma carotenoid concentration. The
color of a soft part, such as the bill, can change rapidly during
any season (Rosenthal et al. 2012), so we could extract both col-
oration and plasma carotenoid level effect sizes from studies of
these ornaments, regardless of molt status.
The means of assessing ornamental coloration is important
to consider in our analysis because color is generally quantified
along one or more of three main axes: hue, or the shade of the
color (e.g., red, orange, yellow); chroma, or the intensity of the
color (also called saturation); and brightness, or the lightness/
darkness of the color. In addition, principle component analy-
sis can be used to create a composite metric directly from the
reflectance spectrum of a color (Montgomerie 2006). Each of
these axes of color tends to relate to different properties of the
colored ornament itself. For example, chroma may be a good
generalization of pigment density, while hue may be more
representative of the proportion of red to yellow pigments in a
carotenoid-colored ornament (Inouye et al. 2001; Hill and
McGraw 2006). The choice of color parameter used in a par-
ticular study is therefore important to include in our analysis
because it may affect study conclusions by representing dif-
ferent properties of the ornament measured.
Statistical Analyses
We performed all analyses using the metafor package (ver. 1.9–
7; Viechtbauer 2010) in R (ver. 3.2.1; R Core Team 2015). We
ran two separate overall meta-analyses, one for plasma carot-
enoid levels and a second for ornamental coloration. Both
analyses used meta-regression to estimate the dose-dependent
effect of carotenoid intake difference (between supplemented
and unsupplemented groups) on response, also including spe-
cies, supplementation duration, and measure type (for color
measurements) as moderators and including a random effect
of study (to control for multiple effect sizes from one experi-
ment). Each model took within-study variation—or the error
around each effect size—into account when estimating overall
effects.
After initial investigation, we discovered that one study in
our plasma carotenoid content analysis (Peters et al. 2011) had
a modest effect size but an order of magnitude larger daily
carotenoid consumption per individual than any other study,
so we ran a separate meta-regression omitting this outlying
data point to better model the patterns in the remaining stud-
ies. We also performed two subgroup analyses for studies of
the plasma content of greenfinches and zebra finches, which
had the greatest number of individual effect sizes (nine and
seven, respectively) and allowed an opportunity to specifically
assess the relationships among parameters in these species.
We also performed a separate analysis of zebra finches for the
relationship of supplementation to coloration. We ran several
furthersubgroupanalysestobetterparsetheeffectsofpar-
ticular model variables when significant sources of hetero-
geneity were held constant (e.g., on data with only hue or only
chroma color parameters). Last, toexamine whether the effects of
supplementation on coloration depend on plasma carotenoid
content, we performed an additional meta-regression to inves-
tigate the effects of species, carotenoid intake difference, color
parameter measured, duration of supplementation, and plasma
effect size on coloration effect size; this analysis was performed
on the subset of studies that measured both plasma carotenoid
content and the color of ornaments.
We investigated the extent of publication bias in the main
plasma carotenoid content and coloration data sets using funnel
plots of effect size versus standard error, a measure of study
precision, according to the guidelines of Koricheva et al. (2013).
Along with a visual examination of plots, we statistically tested
for funnel plot asymmetry using a regression test (Viechtbauer
2010). To estimate the impact of study heterogeneity on meta-
regression results, we calculated Qvalues, which test whether
there was significant residual heterogeneity in effect sizes that
could not be attributed to variation in carotenoid consumption
level and other moderators (Viechtbauer 2010; Koricheva et al.
2013).
Results
Overall, we calculated 40 effect sizes from 15 studies of the
seven focal species. Among the studies we assessed, carotenoid
intake between supplemented and control groups of birds dif-
fered by an average of 19.9 515.1 mg/d/g body mass and ranged
from 0.01 (Stirnemann et al. 2009) to 432.2 mg/d/g body mass
(Peters et al. 2011). Duration of supplementation was similarly
variable, with an average of 39.0 54.70 d and a range from 7
(Karu et al. 2007) to 84 d (Navara and Hill 2003; figs. B1, B2;
app. A; figs. B1–B3 and app. B available online).
Effect of Carotenoid Supplementation on
Plasma Carotenoid Levels
We calculated 21 effect sizes for plasma carotenoid content
from 13 studies of six species (all focal species except the
American goldfinch; figs.1,B1,app.A).Whenweincluded
all data, no variable significantly predicted the effect of sup-
plementation on plasma carotenoid response (all P10.12;
table 1); however, when we omitted one effect size from a study
on great tits that featured an exceptionally large daily carotenoid
intake (Peters et al. 2011), we found that carotenoid intake
difference between supplemented and unsupplemented groups
had a significant effect on plasma carotenoid content response
ratio (table 1). Subgroup models where only greenfinchesorzebra
finches were included also revealed either a trend (greenfinch) or a
000 R.E.Koch,A.E.Wilson,andG.E.Hill
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
significant effect (zebra finch) of carotenoid intake on the re-
sponse of plasma carotenoid levels to supplementation, though
the slope of this relationship differed between the two species:
greenfinches exhibited a larger increase in coloration with in-
creasing carotenoid intake, on average, than zebra finches (ta-
ble 1; fig. 2).
In addition, we found that duration of supplementation had
a significant negative effect on response in the greenfinch sub-
group, indicating that increasing the number of days of supple-
mentation tended to decrease the effect of supplementation on
the response of plasma carotenoid levels in this species (table 1).
This relationship appeared driven by a single study with a 60-d
duration (Peters et al. 2008) and a comparatively low effect size
relative to carotenoid intake difference, so we ran an additional
meta-regression on a data set excluding this data point and
found that the negative effect of supplementation duration was
no longer significant, while the difference in daily carotenoid
intake continued to trend toward significance (Pp0.054; ta-
ble 1). Interestingly, the study of Peters et al. (2008) was ex-
ceptional not only in its long duration but also in that it was the
single study of greenfinch plasma carotenoid levels performed
while the birds were undergoing molt (app. A); if the process of
depositing carotenoids in the growing feathers significantly
altered plasma carotenoid levels, then molt (rather than sup-
plement duration) could be responsible for the lower effect size
relative to carotenoid intake observed in this study.
Significant residual heterogeneity remained in the full data set
model, the full model excluding the Peters et al. (2011) outlier
(described above), and the model including only the zebra finch
data but not in the models containing only greenfinch data (both
with and without Peters et al. 2008; table 1).
Effect of Carotenoid Supplementation on Coloration
We extracted 19 coloration effect sizes from eight studies of five
species: the American goldfinch, diamond firetail, European
greenfinch, Eurasian blackbird, and zebra finch (figs.3,B2;
app. A). Meta-regression indicated that only the type of mea-
surement used to quantify coloration (i.e., hue, chroma, prin-
ciple component analysis) was significant in predicting the mag-
nitude of the effect of supplementation on coloration. While
the presence of supplementation increased coloration in most
studies (fig. 3), neither increasing the difference in carotenoid
intake between supplemented and unsupplemented birds nor
increasing the duration of supplementation affected the dif-
ference in color between experimental and control groups of
birds (table 1). Carotenoid intake continued to have no sig-
nificant relationship with effect size, even in subgroup mod-
els isolating studies measuring only the parameters of hue or
chroma (P10.4), indicating that variation in color measure-
ment was not obscuring effects of variation in carotenoid in-
take in the overall model (table 1).
The separate analysis of zebra finch data also revealed a
significant effect of only measurement type on the response of
coloration to supplementation (table 1). Performing an addi-
tional analysis of the zebra finch data set comprising only effect
sizes measured with hue (excluding one effect size of principle
component analysis; McGraw and Ardia 2003) did not alter the
significance of other model variables; neither days of supple-
mentation nor the magnitude of carotenoid intake had a sig-
nificant effect on the color difference between experimental and
control groups of zebra finches (table 1).
When we examined the relationship between the responses of
coloration and plasma carotenoid content, we found no sig-
nificant effect of any model parameter. Incorporating the
plasma carotenoid content effect size in the coloration model
did reduce the effects of residual heterogeneity from highly
significant in the overall model to nonsignificant (table 1), in-
dicatin g that variation in plasma carotenoid content likely caused
some of the variation in effect sizes present in the overall color-
ation data set. Visual inspection of the plotted relationship
between plasma content and coloration effect sizes revealed
that most points fell below the 1∶1line(fig. 4), so the effect of
supplementationoncolorationtendedtobesmallerthanthe
effect of the same supplementation regimen on plasma carot-
enoid content.
Significant residual heterogeneity remained in the overall
model and the models of only hue or chroma but not in the
zebra finch or plasma carotenoid content models (table 1).
Figure 1. Plasma carotenoid content response ratio (5SE) relative to
thedifferenceincarotenoidintakebetweensupplementedandun-
supplemented groups. Points with no visible error bars represent errors
less than the diameter of the point. The dashed line represents an effect
size of zero, or no difference in plasma carotenoid content between
supplemented and control groups. Species codes are as follows: DIFI,
diamond firetail (Stagonopleura guttata); EUBL, Eurasian blackbird
(Turdus merula); EUGR, European greenfinch (Carduelis chloris); HOFI,
house finch (Haemorhous mexicanus); ZEFI, zebra finch (Taeniopygia
guttata). Not pictured is the effect size from Peters et al. (2011), an out-
lier excluded from main analyses.
Importance of Carotenoid Dose in Songbird Supplementation 000
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
Table 1: Meta-regression model and test of residual heterogeneity results: model effect estimates 5SE
Model and subgroup
No. effect
sizes Intercept
Carotenoid intake
difference
Supplementation
duration Species
Measurement
type
Plasma
effect size
Cochran’s
Q
Plasma carotenoid content:
Overall 21 2.47 5.65*** .0001 5.002 2.0065 5.17 2.28 5.18 NA NA 112.3**
Overall, without outlier 20 2.59 5.58*** .014 5.005*** 2.015 5.016 2.24 5.16 NA NA 117.1***
ZEFI 7 1.01 52.27 .014 5.005*** 2.004 5.05 NA NA NA 46.0***
EUGR 9 1.61 5.18*** .29 5.16* 2.047 5.018*** NA NA NA 4.36
EUGR, without outlier 8 .90 5.55 .44 5.23* 2.03 5.02 NA NA NA 3.33
Coloration:
Overall 19 .12 5.45 .0050 5.011 2.0004 5.0057 .11 5.09 2.14 5.050*** NA 197.0***
ZEFI 7 22.22 51.14* .011 5.012 .001 5.008 NA 1.15 5.39*** NA .9
ZEFI, hue only 6 .098 5.52 .011 5.012 .0013 5.0089 NA NA NA .9
Hue 10 2.038 5.084 .0081 5.010 2.0002 5.0008 .046 5.038 NA NA 149.9***
Chroma 8 .33 5.44 .036 5.081 2.0013 5.0044 2.097 5.19 NA NA 11.3**
Color vs. plasma 11 22.80 51.81 .0016 5.011 .0036 5.011 .55 5.44 .005 5.48 .50 5.47 9.35
Note. NA, not applicable (moderators that were not included in the given model).
*P!0.10.
**P!0.05.
***P!0.01.
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
Publication Bias
Visual inspection of funnel plots indicates little bias in the
studies examined in our analyses (fig. B3), though many effect
sizes were positive; this is not unexpected, given the predicted
physiological relationships between carotenoid intake, plasma
carotenoid content, and coloration. Regression analyses of fun-
nel plot asymmetry indicated no significant bias in either of
the two sets of data (plasma: zp20.83, Pp0.40; color: zp1.03,
Pp0.30).
Discussion
For studies to meaningfully test differential allocation of a limited
pool of carotenoid resources acquired from the diet, they must
provide experimental subjects with biologically relevant caroten-
oid doses. Saturating the diets of birds with carotenoids will
obscure physiological trade-offs that may occur between carot-
enoid absorption, circulation, and use for ornamentation. How-
ever, little justification is generally given for the dosage and ex-
perimental design used in studies that aim to test for carotenoid
trade-offs. To assess the effect that carotenoid dose has on the
physiological responses of birds, we performed meta-regressions
on data extracted from 15 published studies of seven songbird
species. Not surprisingly, and as demonstrated in a previous meta-
analysis (Simons et al. 2012), the presence of carotenoid supple-
mentation tended to increase plasma carotenoid levels and the
expression of carotenoid-based coloration. However, we found
that supplementing an experimental group of birds for a longer
period of time or with a larger dose of carotenoids did not increase
the difference in color between control and supplemented birds.
Because all carotenoids in the system of an adult bird are
derived from the diet, both plasma carotenoid levels and the
expression of carotenoid-based coloration are often assumed
to directly reflect carotenoid intake (Hill et al. 2002; McGraw
2005). The results of our meta-regression of plasma carot-
enoid levels indicate that this assumption is correct across the
range of supplementation doses provided in studies of captive
birds, although the number of days of supplementation did
not affect the response. A dose- but not time-dependent effect
of supplementation on plasma carotenoid response suggests
that the presence of supplementary carotenoids causes an
increase in plasma carotenoid levels to a stable level that varies
according to the dose offered but that does not continue to in-
crease over the duration of the experiment; supplementation ap-
pears to cause the same pattern of increase followed by stabili-
zation in the expression of ornamental coloration, though this
effect is not dose dependent. While it is always important to
validate whether these trends hold true in a particular study
system before applying them to other experimental designs, our
results indicate that future studies need not supplement birds
for long periods of time in order to collect meaningful data on
either plasma carotenoid levels or coloration.
Interestingly, we found a more strongly positive relationship
between increasing supplementation dose and increasing plasma
carotenoid content in greenfinches than in zebra finches. A fun-
damental difference between these two species is that green-
finches have yellow feathers that are colored only during the
annual molt, while z ebra finches have red bills that can be ra pidly
colored or recolored at any time of the year (Hill and McGraw
2006; Rosenthal et al. 2012). The different patterns of carotenoid
absorption and circulation in these two species may reflect the
different physiological requirements for pigmenting feathers
versus bare parts. Specifically, most greenfinches were not un-
dergoing molt at the time of plasma carotenoid content mea-
surement in the studies we examined, so it is possible that they
retained higher levels of ingested carotenoids than zebra finches,
which may have been actively depositing carotenoids in their
bill ornaments at the time of measureme nt. The single data point
Figure 2. Model-fitted plasma carotenoid response ratios relative to the difference in carotenoid intake between supplemented and
unsupplemented groups for two subgroup models comprising data from only zebra finches (ZEFI; circles) or greenfinches (EUGR; triangles).
Lines indicate the model-predicted slope of the response of zebra finches (dashed line) or greenfinches (solid line) to increasing carotenoid
intake, assuming a constant supplementation duration of 25 d. The greenfinch point with the highest value for carotenoid intake difference
represents the effect size from Peters et al. (2011), the only plasma carotenoidcontent measurement for this species that was taken during molt.
Importance of Carotenoid Dose in Songbird Supplementation 000
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
from molting greenfinches (Peters et a l. 2008) showed the lowest
plasma content e ffect size for its given suppl ementation regimen,
which may have been a consequence of the active deposition of
carotenoids into feathers. Unfortunately, the range of studies
available in the literature for our analysis did not have the
breadth required for separate investigations of whether carot-
enoid metabolism t o produce ornamental pigments fro m dietary
pigments (e.g., t o produce red vs. yellow coloratio n) also affected
the relationship between carotenoid intake and plasma content
(Hill and Johnson 2012).
One study of great tits (Peters et al. 2011) had a supple-
mentation dose that was orders of magnitude larger than that
of the other studies included in this analysis; however, the
plasma carotenoid levels measured in this experiment were
within the range of those of other studies. One explanation for
this findingisthatthegreattitsinthisstudymayhavebeenat
the point of maximal carotenoid absorption from their diet
such that even their exceptionally large consumption did not
cause a corresponding increase in circulating carotenoids (the
point of physiological carotenoid saturation). It is also pos-
sible that the insect-rich diet of great tits, as opposed to the
seed-based diet of many of the finch species in our analysis, neces-
sitates corresponding differences in both carotenoid access and
metabolism; however, both the supplemental carotenoid dose
and the plasma effect size of another insect-eating species, the
Eurasian blackbird, was more similar to the finch species in
our study than to these measurements of the great tit. Further
examination of the dose-dependent responses of adult great tits
to carotenoid supplementation as well as measurement of the
quantity of these carotenoids that are allocated to ornamenta-
tion will be essential to extricating how this species makes use
of dietary carotenoids and how it may differ from the cardue-
line finches commonly studied in analyses of carotenoid-based
ornamentation.
In contrast to the strong positive relationship between levels
of carotenoid supplementation and levels of circulating ca-
rotenoids, we found that, while the presence of supplementa-
tion tended to enhance ornamental coloration, increasing the
dose used in supplementation did not cause a corresponding
increase in the response of ornamental coloration. Moreover,
the only significant predictor of how strongly color responded
to supplementation was the parameter used to quantify col-
oration. These results call into question the general and perhaps
overly simplistic assumption that greater carotenoid intake
should inexorably lead to showier coloration. The complexity of
physiological systems involved in carotenoid coloration (Hill
and Johnson 2012) and the links between carotenoid coloration
and metabolism (Johnson and Hill 2013; Hill 2014) make
simple associations between intake and coloration unlikely,
since the expression of coloration is dependent on a variety of
physiological variables beyond carotenoid availability alone. In
fact, our observation that the response of plasma carotenoid
content to supplementation tended to exceed that of coloration
indicates that the levels of carotenoids present in circulation
Figure 4. Ornamental coloration response ratio (5SE) relative to plasma
carotenoid content response ratio (5SE). Symbols with no visible error
bars represent errors less than the diameter of the point. The dashed line
represents a 1∶1 relationshipbetween the two effect sizes. The size of each
symbol represents the magnitude of the carotenoid intake difference
between supplementedand unsupplemented birdsfor the given effect size.
Species codes are as follows: DIFI, diamond firetail; EUBL, Eurasian
blackbird; EUGR, European greenfinch; ZEFI, zebra finch.
Figure 3. Ornamental coloration response ratio (5SE) relative to
carotenoid intake. Symbols with no visible error bars represent errors
less than the diameter of the point. Shading indicates the aspect of
color that was measured. The dashed line represents an effect size of
zero, or no difference in coloration between supplemented and con-
trol groups. Species codes are as follows: AMGO, American goldfinch;
DIFI, diamond firetail; EUBL, Eurasian blackbird; EUGR, European
greenfinch; ZEFI, zebra finch.
000 R.E.Koch,A.E.Wilson,andG.E.Hill
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
were more than adequate for expressing colorful ornaments in
the species examined, so factors other than carotenoid limi-
tation appear responsible for the variation in coloration re-
sponses observed.
Even after accounting for the effects of moderators, many of
our models contained significant residual heterogeneity that
could not always be eliminated in subgroup analyses by species
or by measurement. The persistent variation in effect sizes
within each model emphasizes the unpredictability of response
to supplementation among studies, even within one species and
controlling for variation in carotenoid dose, supplementation
duration, and measurement type. Our results substantiate the
importance of validating that a particular supplementation
regimen is appropriate for a particular experimental design,
perhaps through dosage trials, which are currently used in only
a minority of studies (Alonso-Alvarez et al. 2004; Aguilera and
Amat 2007).
An additional source of variation in our meta-analysis may be
our estimates of carotenoid intake, which are calculated from
predicted food or water intake based on the diet and mass of
each focal passerine species. In fact, while our analyses were
limited to an average measure of consumption for a particular
species, an important consideration for future supplementa-
tion experiments is how food or water intake may vary among
individuals or among treatment groups. The possibility that
birds may use behavioral changes to alter physiological ca-
rotenoid access remains largely unexplored (but see Hill 1995;
McGraw et al. 2003; Peters et al. 2011) and poses a challenge to
detecting internal resource trade-offs. Incorporating measures
of water or food intake with analyses of circulating carotenoids
and ornamental coloration is a simple but highly valuable step
to understand the true magnitude—and, consequently, bio-
logical relevance—of supplementation.
Despite the large number of studies that have tested the phys-
iological effects of carotenoids on ornamentation, only a small
sample of studies performed controlled supplementation of adult
birds with carotenoid-based coloration. Although this small sam-
ple size necessarily limits the breadth of the inferences that can
be drawn from our study, we found some intriguing patterns that
are not necessarily intuitive. Our ultimate goal is to emphasize
important methodological and theoretical considerations for fu-
ture studies using carotenoid supplementation to assess the con-
dition dependence of carotenoid-based ornaments. Improving the
clarity of the relationships between carotenoid intake, circulation,
and deposition in ornamentation in a variety of species will be an
important step to better understanding the size and function of the
pool of dietary carotenoids available to songbirds and may reduce
ambiguity in the results of studies searching for carotenoid allo-
cation trade-offs.
Acknowledgments
The National Science Foundation Graduate Research Fellow-
ship Program provided financial support for R.E.K. during
data collection and manuscript preparation. We would also like
to thank Carlos Alonso-Alvarez, Wendy Hood, and Kevin
McGraw for sharing unpublished experimental data for anal-
ysis and three anonymous reviewers for feedback on the man-
uscript.
Literature Cited
Aguilera E. and J.A. Amat. 2007. Carotenoids, immune response
and the expression of sexual ornaments in male greenfinches
(Carduelis chloris). Naturwissenschaften 94:895–902.
Alonso-AlvarezC.,S.Bertrand,G.Devevey,M.Gaillard,J.
Prost, B. Faivre, and G. Sorci. 2004. An experimental test of
the dose-dependent effect of carotenoids and immune ac-
tivation on sexual signals and antioxidant activity. Am Nat
164:651–659.
Alonso-Alvarez C., L. Perez-Rodriguez, R. Mateo, O. Chastel,
and J. Vinuela. 2008. The oxidation handicap hypothesis and
the carotenoid allocation trade-off. J Evol Biol 21:1789–1797.
Baeta R., B. Faivre, S. Motreuil, M. Gaillard, and J. Moreau.
2008. Carotenoid trade-off between parasitic resistance and
sexual display: an experimental study in the blackbird
(Turdus merula). Proc R Soc B 275:427–434.
Blount J.D., N.B. Metcalfe, T.R. Birkhead, and P.F. Surai. 2003.
Carotenoid modulation of immune function and sexual attrac-
tiveness in zebra finches. Science 300:125–127.
Butler M.W. and K.J. McGraw. 2013. Immune function is
related to adult carotenoid and bile pigment levels, but not
to dietary carotenoid access during development, in female
mallard ducks. J Exp Biol 216:2632–2640.
Calder W. and E.J. Braun. 1983. Scaling of osmotic regulation
in mammals and birds. Am J Physiol 244:R601–R606.
Caraco T., S. Martindale, and T.S. Whittam. 1980. An em-
pirical demonstration of risk-sensitive foraging preferences.
Anim Behav 28:820–830.
CostantiniD.,C.Coluzza,A.Fanfani,andG.Dell’Omo. 2007.
Effects of carotenoid supplementation on colour expres-
sion, oxidative stress and body mass in rehabilitated captive
adult kestrels (Falco tinnunculus). J Comp Physiol B 177:
723–731.
Costantini D. and A.P. Møller. 2008. Carotenoids are minor
antioxidants for birds. Funct Ecol 22:367–370.
del Hoyo J., A. Elliott, J. Sargatal, and J. Cabot. 2010.
Handbook of the birds of the world. Lynx, Barcelona.
Goodwin T.W. 1984. The biochemistry of the carotenoids.
Springer, Berlin.
Hadfield J.D. and I.P.F. Owens. 2006. Strong environmental
determination of a carotenoid-based plumage trait is not
mediated by carotenoid availability. J Evol Biol 19:1104–
1114.
HartleyR.C.andM.W.Kennedy.2004.Arecarotenoidsared
herringinsexualdisplay?TrendsEcolEvol19:353–354.
Hill G.E. 1994. House finches are what they eat: a reply to
Hudon. Auk 111:221–225.
———. 1995. Seasonal variation in circulating carotenoid
pigments in the house finch. Auk 112:1057–1061.
Importance of Carotenoid Dose in Songbird Supplementation 000
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
———. 1996. Redness as a measure of the production cost of
ornamental coloration. Ethol Ecol Evol 8:157–175.
———. 1999. Is there an immunological cost to carotenoid-
based ornamental coloration? Am Nat 154:589–595.
———. 2002. A red bird in a brown bag: the function and
evolution of colorful plumage in the house finch. Oxford
University Press, Oxford.
———. 2006. Environmental regulation of ornamental col-
oration. Pp. 507–560inG.E.HillandK.J.McGraw,eds.
Bird coloration: mechanisms and measurements. Harvard
University Press, Cambridge, MA.
———. 2011. Condition-dependent traits as signals of the
functionality of vital cellular processes. Ecol Lett 14:625–
634.
———. 2014. Cellular respiration: the nexus of stress, con-
dition, and ornamentation. Integr Comp Biol 54:645–657.
Hill G.E., C.Y. Inouye, and R. Montgomerie. 2002. Dietary
carotenoids predict plumage coloration in wild house finches.
Proc R Soc B 269:1119–1124.
Hill G.E. and J.D. Johnson. 2012. The vitamin A–redox hy-
pothesis: a biochemical basis for honest signaling via ca-
rotenoid pigmentation. Am Nat 180:E127–E150.
Hill G.E. and K.J. McGraw. 2006. Bird coloration: mechanisms
and measurements. Harvard University Press, Cambridge,
MA.
HõrakP.,L.Saks,I.Ots,T.Kullissaar,H.Kollist,andM.
Zilmer. 2003. Physiological effects of immune challenge in
captive greenfinches (Carduelis chloris). Can J Zool 81:371–
379.
HõrakP.,L.Saks,M.Zilmer,U.Karu,andK.Zilmer.2007.Do
dietary antioxidants alleviate the cost of immune activa-
tion? an experiment with greenfinches. Am Nat 170:625–
635.
HudonJ.1994.Showiness,carotenoids,andcaptivity:acom-
ment on Hill (1992). Auk 111:218–221.
Inouye C.Y., G.E. Hill, R.D. Stradi, and R. Montgomerie. 2001.
Carotenoid pigments in male house finch plumage in re-
lation to age, subspecies, and ornamental coloration. Auk
118:900–915.
Johnson J.D. and G.E. Hill. 2013. Is carotenoid ornamentation
linked to the inner mitochondria membrane potential? a
hypothesis for the maintenance of signal honesty. Biochimie
95:436–444.
KaruU.,L.Saks,andP.Horak.2007.Carotenoidcolorationin
greenfinches is individually consistent irrespective of for-
aging ability. Physiol Biochem Zool 80:663–670.
Koricheva J., J. Gurevitch, and K. Mengersen. 2013. Handbook
of meta-analysis in ecology and evolution. Princeton Uni-
versity Press, Princeton, NJ.
McGraw K. 2005. Interspecific variation in dietary carotenoid
assimilation in birds: links to phylogeny and color orna-
mentation. Comp Biochem Physiol B 142:245–250.
McGrawK.J.andD.R.Ardia.2003.Carotenoids,immuno-
competence, and the information content of sexual colors:
an experimental test. Am Nat 162:704–712.
McGraw K.J., O.L. Crino, W. Medina-Jerez, and P.M. Nolan.
2006. Effect of dietary carotenoid supplementation on food
intake and immune function in a songbird with no carot-
enoid coloration. Ethology 112:1209–1216.
McGrawK.J.,A.J.Gregory,R.S.Parker,E.Adkins-Regan,and
R.Prum.2003.Diet,plasmacarotenoids,andsexualcol-
orationinthezebrafinch (Taeniopygia guttata). Auk 120:
400–410.
McGrawK.J.,G.E.Hill,R.Stradi,andR.S.Parker.2001.The
influence of carotenoid acquisition and utilization on the
maintenance of species-typical plumage pigmentation in
male American goldfinches (Carduelis tristis) and northern
cardinals (Cardinalis cardinalis). Physiol Biochem Zool 74:
843–852.
———. 2002. The effect of dietary carotenoid access on sexual
dichromatism and plumage pigment composition in the
American goldfinch. Comp Biochem Physiol B 131:261–269.
McGraw K.J. and K.C. Klasing. 2006. Carotenoids, immunity,
and integumentary coloration in red junglefowl (Gallus gallus).
Auk 123:1161–1171.
MøllerA.P.,C.Biard,J.D.Blount,D.C.Houston,P.Ninni,N.
Saino, and P.F. Surai. 2000. Carotenoid-dependent signals:
indicators of foraging efficiency, immunocompetence or
detoxification ability? Avian Poult Biol Rev 11:137–159.
Montgomerie R. 2006. Analyzing colors. Pp. 90–147 in G.E.
Hill and K.J. McGraw, eds. Bird coloration: mechanisms
and measurements. Harvard University Press, Cambridge,
MA.
Mougeot F., J. Martinez-Padilla, J.D. Blount, L. Perez-
Rodriguez, L.M.I. Webster, and S.B. Piertney. 2010. Oxi-
dative stress and the effect of parasites on a carotenoid-
based ornament. J Exp Biol 213:400–407.
NagyK.A.,I.A.Girard,andT.K.Brown.1999.Energeticsof
free-ranging mammals, reptiles, and birds. Annu Rev Nutr
19:247–277.
Navara K.J. and G.E. Hill. 2003. Dietary carotenoid pigments
and immune function ina songbird with extensive carotenoid-
based plumage coloration. Behav Ecol 14:909–916.
Pérez-Rodríguez L., F. Mougeot, and C. Alonso-Alvarez. 2010.
Carotenoid-based coloration predicts resistance to oxida-
tive damage during immune challenge. J Exp Biol 213:1685–
1690.
Peters A., K. Delhey, S. Andersson, H. van Noordwijk, and
M.I. Foerschler. 2008. Condition-dependence of multiple
carotenoid-based plumage traits: an experimental study.
Funct Ecol 22:831–839.
PetersA.,S.Magdeburg,andK.Delhey.2011.Thecarotenoid
conundrum: improved nutrition boosts plasma carotenoid
levels but not immune benefits of carotenoid supplemen-
tation. Oecologia 166:35–43.
Rasband W.S. 1997–2014. ImageJ. National Institutes of Health,
Bethesda, MD.
R Development Core Team. 2015. R: a language and envi-
ronment for statistical computing. R Foundation for Sta-
tistical Computing, Vienna.
000 R.E.Koch,A.E.Wilson,andG.E.Hill
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
RohatgiA.2013.WebPlotDigitizer.http://arohatgi.info/Web
PlotDigitizer.
Rosenthal M.F., T.G. Murphy, N. Darling, and K.A. Tarvin.
2012. Ornamental bill color rapidly signals changing con-
dition. J Avian Biol 43:553–564.
Sepp T., U. Karu, E. Sild, M. Manniste, and P. Horak. 2011.
Effects of carotenoids, immune activation and immune sup-
pression on the intensity of chronic coccidiosis in greenfinches.
Exp Parasitol 127:651–657.
SildE.,T.Sepp,M.Manniste,andP.Horak.2011.Carotenoid
intake does not affect immune-stimulated oxidative burst in
greenfinches. J Exp Biol 214:3467–3473.
SimonsM.J.,A.A.Cohen,andS.Verhulst.2012.Whatdoes
carotenoid-dependent coloration tell? plasma carotenoid
level signals immunocompetence and oxidative stress state
in birds: a meta-analysis. PLoS One 7:e43088.
SmithH.G.,L.Raberg,T.Ohlsson,M.Granbom,andD.
Hasselquist. 2007. Carotenoid and protein supplementation
have differential effects on pheasant ornamentation and
immunity. J Evol Biol 20:310–319.
Stirnemann I., G. Johnston, B. Rich, J. Robertson, and S.
Kleindorfer. 2009. Phytohaemagglutinin (PHA) response and
bill-hue wavelength increase with carotenoid supplementation
in diamond firetails (Stagonopleura guttata). Emu109:344–351.
Svensson P.A. and B.B.M. Wong. 2011. Carotenoid-based signals
in behavioural ecology: a review. Behaviour 148:131–189.
Tella J.L., J. Figuerola, J.J. Negro, G. Blanco, R. Rodriguez-
Estrella, M.G. Forero, M.C. Blazquez, A.J. Green, and F.
Hiraldo. 2004. Ecological, morphological and phylogenetic
correlates of interspecific variation in plasma carotenoid
concentration in birds. J Evol Biol 17:156–164.
Viechtbauer W. 2010. Conducting meta-analyses in R with the
metafor package. J Stat Softw 36:1–48.
vonSchantzT.,S.Bensch,M.Grahn,D.Hasselquist,andH.
Wittzell. 1999. Good genes, oxidative stress and condition-
dependent sexual signals. Proc R Soc B 266:1–12.
Importance of Carotenoid Dose in Songbird Supplementation 000
This content downloaded from 23.235.32.0 on Mon, 23 Nov 2015 11:09:41 AM
All use subject to JSTOR Terms and Conditions
